. 2021 Oct 1;142(11):3349–3366. doi: 10.1007/s00402-021-04163-w

Table 2.

Methods and results of the included studies

Study	Rotation angles, Flexion angles, further information	Main findings, conclusions, main limitations
Brouwer et al. [8]	Rotation: from 30° ER to 30° IR (15° increments) Flexion: 0°, 15°, 30°	Rotation or flexion alone did not change the HKA (-10° in NR) by more than 1°. Rotation and flexion led to large changes of the HKA: HKA -10° for 15° flexion with 15° IR, HKA -14° for 15° flexion with 15° ER; HKA 2° for 15° flexion with 30° IR, and HKA -16° for 15° flexion with 30° ER. For the varus aligned limb, flexion in combination with ER caused more changes than flexion with IR Limitations: only one specimen was analyzed
Hunt et al. [19]	Rotation: 0°, 15° IR and 15° ER of the foot	IR of the foot resulted in less measured varus alignment and ER leads to greater measured varus alignment: HKA: -7.2° ± 3.5° in 15° ER, -6.1° ± 3.8° in NR, -3.6° ± 2.7° in 15° IR; MAD: 27.5 mm ± 13.8 in 15° ER, 22.9 mm ± 14.0 in NR, 14.3 mm ± 9.8 in 15° IR Limitations: rotation is controlled by the foot position – not by the position of the patella; no information was given about the deformity in the axial or sagittal plane
Jamali et al. [21]	Rotation: from 12° ER to 12° IR (3° increments) Others: 3D models were created from the CT data. A.p. images were obtained using a rendering software	Small amount of rotation led to a statistically significant difference relative to the neutral position for all mechanical and anatomical alignment parameters except for aLDFA and mLDFA. Effects may be small and their clinical importance is unknown. For the HKA, TFA and AMA 3° rotation led to a statistically significant difference compared to the NR position. IR leads to more valgus and ER leads to more varus. AMA was 5.33° in NR, 5.4° in 12° IR, and 5.1° in 12° ER (average decrease of 0.0146° per degree of rotation) Limitations: Exact values of the change of alignment parameters were not reported; for comparison to other studies, the change of TFA and HKA due to limb rotation was extracted form a graph
Jud et al. [24]	Rotation: from 30° ER to 30° IR (10° increments) Flexion: 0°, 5°, 10°, 15°, 20°, 30° Others: For each simulation a.p. projections were calculated from the 3D surface model, and the HKA were measured	HKA is not influenced by rotation when the knee is in 0° flexion, but it is by the combination of rotation and flexion. Changes of the HKA are comparable between patients with coronal alignment between 9° varus and 9° valgus: Rotation and 0° flexion—HKA: Median change referenced to the HKA-baseline 0° (min./max. deviation from the median: -0.8°/0.2°) in 20° ER, 0° (-0.3°/0.0°) in 10° ER 0.1° (-0.3°/0.2°) in 10° IR, 0.2° (-0.9° /0.7°) in 20° IR. Rotation and 10° flexion – HKA: -2.6° (-1.2°/0.3°) in 20° ER, -1.1° (-0.6°/0.2°) in 10° ER, 0.6 (-0.3/0.1) in NR, in 2.3° (-0.4°/0.4°) 10° IR, 3.9° (-0.9°/0.9°) in 20° IR Limitations: All patients had normal sagittal alignment and femoral torsion; study is based on non-weight-bearing computer simulations
Kannan et al. [25]	Rotation: from 0° to 20° ER (5° increments) Flexion: 0°, 5°, 10°, 15°, 20°	HKA did not vary > 1° with knee flexion or rotation alone. The combination of both factors can alter the apparent HKA substantially: 10° ER, 5° flexion change of 2°; 10° ER, 10° flexion change of 2°; 20° ER, 10° flexion change of 4°. MPTA and FFA changed by ≤ 1° by flexion or rotation and ≤ 3° in combination of rotation and flexion Limitations: LLR were not performed in accordance to Paley; IR was not analyzed; only one synthetic sample was analyzed
Kawahara et al. [26]	Rotation: 0°, 5° ER and IR, 10° ER and IR Others: digitally reconstructed radiography images parallel to the surgical epicondylar axis (neutral rotation; NR), 5° and 10° external rotation (ER) or internal rotation (IR) were reconstructed from preoperative CT. Based on the images, open-wedge HTO was planned aiming for a postoperative WBL of 62.5%. The planned opening angle were measured in each image	The preoperative knee extension angle was − 2.4° ± 3.8°. Weight-bearing line (WBL) changed from 25.6 ± 10.5% in 10°ER, 26.8 ± 9.5% in NR, to 28.5 ± 9.2% in 10°IR. Differences of the WBL between the measurement at 10°, 5° ER, 5°, 10°IR and the neutral position significantly correlated with the knee flexion angle The analysis of the effect of limb malrotation on preoperative planning of open-wedge HTO showed that the weight-bearing line increased and the medial proximal tibial angle gradually increased from ER to IR, whereas the opening angle gradually decreased. As the opening angle changed only within 0.5° on average, authors concluded that malrotation of the LLR by < 10° does not influence preoperative planning in open-wedge HTO in cases without knee extension deficits. However, knees with > 10° knee extension deficit showed a difference in the measured open angle of > 1° due to rotation Limitations: Maximum and minimum values of the measured difference were not given. Alignment and flexion deficits were measured using non-weight-bearing CT imaging
Kawakami et al. [27]	Rotation: measured on LLR by superimposing 3D bone models Flexion: obtained from a lateral view and calculated by the computer software Others: LLR and 3D Models were obtained using a CT. A computer software was developed to calculate TFA and HKA, its changes due to rotation, to measure knee flexion and simulate knee osteotomy from the 3D bone model. Closed-wedge HTO was simulated with a target of 4° HKA postoperative	The mean limb rotation on radiographs was 7.4 ± 3.9° of IR (range: 8° ER to 14° IR). The mean TFA and HKA on a.p. standing radiographs was 183.5 ± 5.5° (range: 178–205) and -9.2 ± 5.6° (range: -30 to -4). The mean change of HKA was 1.6° ± 1.3° (0.2–4.9) and of the TFA 3.5° ± 2.2° (0.4–8.6°). No significant correlation was found between frontal alignment on AP radiographs and the change in alignment with limb rotation. The change in HKA and FTA with limb rotation increased as the knee flexion increased Rotation may affect measurement of lower limb alignment for knee osteotomy: With IR and ER by 5°, 10°, and 15° of the osteotomy plane, the FTA changed by 0.2 ± 0.1°, 0.4 ± 0.3°, and 0.8 ± 0.4° and the HKA changed by 0.2 ± 0.1°, 0.4 ± 0.2°, and 0.7 ± 0.3°. HKA and TFA after virtual closed-wedge tibial osteotomy were measured of a case with a rotational error of the osteotomy plane. The proximal cutting surface was reattached to the distal cutting surface. The mean change in TFA was 1.6 ± 0.6°, 3.2 ± 1.1°, and 4.8 ± 1.7° for 5°, 10°, and 15° IR and ER errors in reattaching the osteotomy plane. The mean change in HKA was 0.9° ± 0.5°, 1.9 ± 1.0°, and 2.8 ± 1.5° with an internal and external 5°, 10°, and 15° osteotomy plane rotation Limitations: transverse images were obtained using a helical CT scanner while the patient was in supine position; only varus knees were included
Kendoff et al. [28]	Rotation: 0°, 5° ER and IR, 10° ER and IR, maximal rotation Flexion: 0°, 5°, 10°, 20° Others: The mechanical leg axis and the axis deviations were assessed using a navigation system	In full extension, axis measurement deviations ranging from 0.4° to 4.3° with larger deviations occurring with increasing rotation (IR or ER): mechanical axis deviation with full extension 10° IR 1.5°, 10° ER 2.1°, maximal IR 4.3°, and maximal ER 3.9°. When the knee was flexed, larger deviations were found: 5° flexion with 5°IR led to alignment errors of 3.4°, and with 5°ER to 2.8°. Over- and undercorrection might occur Limitations: measurement deviation was measured, not the absolute mechanical axis; coronal plane alignment was measured with a navigation system
Khare and Jaramaz [29]	Rotation: 0°, 10° IR, 10° ER Others: Digitally reconstructed radiographs were obtained from CT scans. Landmarks were manually identified in the radiographs and the 3D CT scans were compared and variations of the positions were measured	Rotation leads to increased errors in certain landmarks (as large as 13.1 mm for the femoral knee center and 13.6 mm for the lateral malleolus). Despite of a large variation in the measurement of landmark points, the estimation of tibial and femoral mechanical axes did not suffer inaccuracy (maximum error of 1.46° for the femoral mechanical axis and 0.66° for the tibial mechanical axis) Limitations: No weight-bearing condition and radiographs were obtained by digitalized radiographs from a CT scan
Lee et al. [33]	Rotation: 0°, 15° IR an 30° ER of the foot	Foot position in ER shows less varus and IR shows more varus. The weight bearing line shifted laterally in the 30° position and shifted medial in the 15° internal position compared to the neutral position (1.8 mm lateral, 0.2 mm medial in the LLR, and 3.5 mm lateral and 3 mm medial in the local radiograph). Results of the ratio of weight-bearing line/tibial plateau width were similar to those of the weight-bearing line Limitations: the limb was positioned by the foot position; only standard deviations were presented to report the variation from the mean
Lonner et al. [35]	Rotation: from 20° ER to 25° IR (5° increments) Flexion: 0°, 10°	ER causes less apparent anatomic valgus and IR pretends more valgus. Even in a well-aligned TKA, limb positioning will alter alignment measurements, making objective evaluation difficult: Rotation and 0° Flexion: 2.6° TFA in 20° ER, 5.7° in 25° IR; Rotation and 10° Flexion: 2.3° in 20° ER, 6.7° in 25° IR. Tibial alignment changed with 10° flexion ranging from -5° varus in 20°ER to 3° valgus in 25°IR Limitations: mechanical alignment was not measured; non-weight-bearing condition; short-plane radiographs; for comparison to other studies, the change of TFA due to limb rotation was extracted from a graph
Maderbacher et al. [36]	Rotation: rotation errors were calculated by the tibial-fibular overlap using a formula [38] Flexion: for all radiographs full extension and 10° flexion were assumed Others: The impact of observed rotational errors was calculated using data of previous studies: for each degree of rotation, the HKA changed in extended knees by 0.0697° and with 10° flexion by 0.0986; femoral and tibial component alignment changed by 0.0824 and 0.0504. Malalignment was defined as a deviation of femoral or tibial component alignment > 2° from 90° or deviation > 3° from neutral HKA (0°)	Limbs were positioned in average in 8.1° ± 9.3° IR (range: 36° IR and 16° ER). Mean measurement errors due to rotation were quite small. However, large ranges of wrong measurements were found indicating significant and clinically relevant measurement errors in some radiographs Mean differences of HKA between measured alignment before and after a mathematical rotational correction were -0.6° ± 0.6° ranging from -2.5° to 1.1° (0° knee flexion) and -0.8° ± 0.9° ranging from − 3.5° to 1.6° (10° knee flexion). 11 out of 100 patients were wrongly assigned to either mal- or well-aligned, which was defined as HKA within ± 3° varus or valgus Limitations: see below (Maderbacher et al. [37])
Maderbacher et al. [37]	Rotation: rotation errors were calculated by the fibular overlap using a formula [38] Flexion: for all radiographs full extension and 10° flexion was assumed Others: The impact of observed rotational errors on alignment was calculated using data of previous studies: for each degree of rotation AMA changed by 0.0558, mLPFA by 0.1892, mLDFA by 0.0824, mMPTA by 0.0504, mLDTA by 0.2185, HKA in extended knees by 0.0697° and with 10° flexion by 0.0986	Malrotation is regularly present in LLR with average rotation of the limb by -8.0 ± 9.0° IR (range: 29.4° IR and 22.1° ER). Measured alignment parameters before and after rotational correction showed the following mean differences: HKA in full knee extension: 0.6° ± 0.6° (-1.5° to 2.1°); HKA in 10° knee flexion 0.8° ± 0.9° (-2.2° to 2.9°); AMA 0.4° ± 0.5° (-1.2° to 1.6°); mLPFA 1.5° ± 1.7° (-4.2° to 5.6°); mMPTA 0.4° ± 0.5° (-1.1° to 1.5°); mLDFA 0.7° ± 0.7° (-1.8° to 2.4°); mLDTA 1.7° ± 2.0° (-4.8° to 6.4°). As all measured parameters are influenced by malposition, correct limb rotation needs to be verified Limitations: The formula used to calculate the rotational error can predict lower limb rotation on LLR with a SD of 5°, but with IR the accuracy of calculated rotations decreases; calculated corrections are based on previous studies by Lonner, Jiang and Insall, and Radtke; these values are abstract and vary between mentioned studies. Moreover, these studies were conducted, as outlined by the present review, with small sample sizes; the study did not provide new data regarding the relationship between rotation and alignment deviation
Meijer et al. [40]	Rotation: from 20° ER to 20° IR Flexion: 0°, 5°, 10°, 15°, 20° Others: change of the tibiofemoral alignment from 15° varus to 15° valgus (5° increments)	Changes of rotation, flexion, and tibiofemoral alignment in the frontal plane alone had no major impact on the measured HKA (different rotations with 0° varus/valgus and 0° flexion changed between 0.1 and 1.5°). Changes in varus/valgus alignment and rotation with a fully extended knee did not show major differences in HKA (different rotations with 0° flexion and 10° valgus and 10° varus led to a change of HKA by 1.6°–2.1° and -1.7°–0.3). Flexion (20°) combined with rotation with a neutral alignment showed a HKA variation of up to 13.8° (range: -5.2 to 8.6°), a change of rotation and 10° flexion in a neural aligned knee led to a change of HKA between -2.3 and 4.2°. Combining all three factors (15° valgus, 20° flexion) led to maximum HKA variation of up to 16.5° (-10.5–6°). A change of rotation with 10° varus and 10° flexion led to different measured HKA between -2.0–4.9° Limitations: some values of the change of the alignment parameters were not reported; for comparison to other studies, the change of HKA due to limb rotation was extracted form graphs
Moon et al. [42]	Rotation: was assessed by measuring the deviation of the patellar center inward or outward relative to the midpoint between the femoral condyles Flexion: the knee extension and flexion was measured on the EOS images Others: HKA measurement between LLR and EOS was calculated (dHKA) Patients were divided into two subgroups on the sagittal plane (knee flexion and extension group) and into two subgroups based on the axial plane (knee internal and external rotation)	The mean value of the HKA on LLR was 1.2 ± 3.9°, knee flexion angle 5.3 ± 5.3°, and patellar rotation of 4.6 ± 4.0%. Significant correlation was found between dHKA and degree of knee flexion/extension (r = 0.368). No significant correlation was found between dHKA and the axial limb rotation. A significant linear relationship between dHKA and the knee flexion/extension in patients with a patellar rotation more than 3% was found (r² = 0.257) Authors concluded that the measurement accuracy of coronal alignment on LLR was influences by knee flexion, which was significant in case of small rotation. Therefore, it would be important to check the patella position especially in patients with knee flexion Limitations: correlations were presented, but no exact values to describe the relationship between rotation and measured HKA; rotation was measured with the patella position
Radtke et al. [47]	Rotation: from 20° ER to 20° IR (5° increments) Flexion: full extension	Significant effects of rotation on the measured alignment were found: IR simulated tibiofemoral valgus and ER a varus alignment. The alignment parameters in NR were AMA 6.2°, mLPFA 97.5°, mLDFA 88.2°, mMPTA 89.4°, mLDTA 94°. From 20° IR to 20° ER AMA changed from 6.8° to 4.6°, mLPFA from 101.6° to 93.6°, mLDFA 90.6° to 86.8°, mMPTA 90.4° to 88.5°, and mLDTA from 98.9° to 90.5° Limitations: only one synthetic femur and tibia; HKA or TFA was not measured
Sabharwal et al. [49]	Rotation: gauged as the percentage of proximal fibular overlap and calculated as the percentage of the proximal fibular width that was overlapped by the adjacent tibial metaphyses, at the level of the proximal fibular physis. A large value indicates greater ER Others: 2 LLR were conducted: one before and the other after ring fixator removal. Difference in the frontal alignment measurement and the limb rotation was measured	The limb tended to be positioned in ER when the radiograph was taken with the circular fixator still attached to the extremity. This tendency for ER of the limb likely contributed to an overestimation of varus alignment. Absolute difference of the MAD between the 2 LLR was 11.5 mm for the treated limb (95%CI with fixator: -3.3 (valgus) to 15.7 mm (varus)); 95%CI without fixator: -6.9 to 10 mm). The mean absolute difference in the fibular overlap between the 2 LLR was 21% (95%CI with fixator: 24% to 48%; 95%CI without fixator: 16% to 33%). Linear regression analysis demonstrated that with an increase of fibular overlap, there was a progressive increase in the magnitude of discrepancy in the measurement of MAD between the 2 LLR Limitations: the rotation was approximated by the percentage of the proximal fibular overlap
Stricker and Faustgen [55]	Rotation: 0°, 15° ER, 30° ER	Limb malrotation can reduce the validity of HKA and TFA progressively. HKA: -27.7° in NR and increased by 2.4° in 15° ER and 3.0° in 30° ER; TFA: -24.8° in NR and increased by 3.0° in 15° ER and 6.7° in 30° ER. For MDA (average 12.8° in 0° rotation according to Levine-Drennan) unpredictable variability was noted during ER, which can lead to misdiagnosis and mistreatment of Blount’s disease Limitations: the limb positioning was controlled with ultrasound; only standard deviations were presented to report the variation from the mean
Swanson et al. [56]	Rotation: from 20° ER to 20° IR (10° increments) Others: Three deformed limbs: A 18° valgus, B 7° varus and C 7° valgus (TFA). Each model had 5 radiographs in each position With a graphic software, high tibial osteotomies were simulated for model B. Osteotomies aimed to achieve 8° valgus	The effect of rotation on measured TFA was more sensitive in varus and valgus deformed limbs than in a neutral limb. Severe valgus deformities are more sensitive to the effects of ER; severe varus deformities are more sensitive to effects of IR. TFA of model A: 13.6° in 20° ER, 16.7° in 10° ER, 18.7° valgus in NR, 20.2° in 10° IR, 21.4° in 20° IR. Model B: -7.8° in 20° ER, -7.4° in 10° ER, -6.4° in NR., -5.3° in 10° IR, -3.7° in 20° IR. Model C: 4.6° in 20° ER, 6.3° in 10° ER, 6.3° in NR, 6.9° in 10° IR, 7.3° in 20° IR. AMA varied less than 1° in all three models between 20° IR and ER A simulated osteotomy based on the measured TFA of model B in 20° IR an undercorrection occurred. Ideal correction of a varus deformity was achieved when the osteotomy was simulated based on LRR with the limb in the neutral position. Based on the measured TFA in 20°ER, an overcorrection was observed Limitations: axial rotation was assessed with the help of a goniometer
Thelen et al. [58]	Rotation: from 20° ER to 20° IR (5° increments) Flexion: 1 bone in 0°, 1 bone in 9°, 1 bone in 18° Others: Biplanar slot scanning system (EOS) took an a.p. and a lateral radiograph of the models (5° valgus) for each rotation position	The impact of rotation on the HKA is relatively small in the absence of sagittal knee angulation. An axial rotation of 10° (IR or ER) resulted in a measurement error (defined as difference between the measured HKA and the actual model alignment of 5°) in the 2D a.p. view of 0.4°, 1.9°, and 3.1° and an axial rotation of 20° (IR or ER) resulted in an average measurement error of 1.4°, 4.7°, and 6.8° with 0°, 9°, and 18° of knee flexion. Summarizing the measurements of 3 observers, HKA varied from 3.8°-6.4° (model 0° flexion), 0.3–9.7° (model 9° flexion) and -0.7 – 11.8° (model 18° flexion) Limitations: only standard deviations were presented to report the variation from the mean; for comparison to other studies, the change of HKA due to limb rotation was extracted form a graph
Wright et al. [63]	Rotation: from 20° ER to 20° IR (10° increments) Flexion: 0°, 20°, 30°	Little overall difference between the rotation groups were found. The reliability of lower limb alignment measured from LLR is satisfactory. The radiographer positioned the limb within 3.1° ± 2.7° from the neutral position Rotation and 0° Flexion of the first limb: 2.0° in 20° ER, 1.7° in 10° ER, 2.0° in NR, 2.3° in 10° IR, 2.0° in 20° IR. Rotation and 0° Flexion of the second limb: 7.0° in 20° ER, 3.3° in 10° ER, 1.0° in NR, -0.7° in 10° IR, -2.0° in 20° IR. Rotation and 20° Flexion—TFA: 2.0° in 20° ER, 2.0° in NR, 0.7° in 20° IR of the first limb; 2.7° in 20° ER, 4.6° in NR, 4.0° in 20° IR of the second limb Limitations: different findings between the limbs; only two knee amputates were analyzed; only anatomical measurements were analyzed
Yoo et al. [64]	Rotation: assessed by measuring the fibula/tibia overlap, as well as the tibial and femoral component rotation from AP radiographs conducted by EOS imaging system Flexion: Flexion was measured from lateral EOS radiographs Others: correlations referring to the HKA angle were calculated	There was a significant correlation for the fibular overlap (r = 0.292)/the femoral component rotation (r = 0.317) and the HKA angle. Moreover, combined rotation and knee flexion had a greater effect on the coronal alignment than flexion/rotation alone. This was more frequently observed during the early period after TKA Limitations: Rotation was only described by the fibular overlap and the component position. Relationship between rotation/flexion was expressed by calculating correlation coefficients

Study

Rotation angles, Flexion angles, further information

Main findings, conclusions, main limitations

Brouwer et al. [8]

Rotation: from 30° ER to 30° IR (15° increments)

Flexion: 0°, 15°, 30°

Rotation or flexion alone did not change the HKA (-10° in NR) by more than 1°. Rotation and flexion led to large changes of the HKA: HKA -10° for 15° flexion with 15° IR, HKA -14° for 15° flexion with 15° ER; HKA 2° for 15° flexion with 30° IR, and HKA -16° for 15° flexion with 30° ER. For the varus aligned limb, flexion in combination with ER caused more changes than flexion with IR

Limitations: only one specimen was analyzed

Hunt et al. [19]

Rotation: 0°, 15° IR and 15° ER of the foot

IR of the foot resulted in less measured varus alignment and ER leads to greater measured varus alignment: HKA: -7.2° ± 3.5° in 15° ER, -6.1° ± 3.8° in NR, -3.6° ± 2.7° in 15° IR; MAD: 27.5 mm ± 13.8 in 15° ER, 22.9 mm ± 14.0 in NR, 14.3 mm ± 9.8 in 15° IR

Limitations: rotation is controlled by the foot position – not by the position of the patella; no information was given about the deformity in the axial or sagittal plane

Jamali et al. [21]

Rotation: from 12° ER to 12° IR (3° increments)

Others: 3D models were created from the CT data. A.p. images were obtained using a rendering software

Small amount of rotation led to a statistically significant difference relative to the neutral position for all mechanical and anatomical alignment parameters except for aLDFA and mLDFA. Effects may be small and their clinical importance is unknown. For the HKA, TFA and AMA 3° rotation led to a statistically significant difference compared to the NR position. IR leads to more valgus and ER leads to more varus. AMA was 5.33° in NR, 5.4° in 12° IR, and 5.1° in 12° ER (average decrease of 0.0146° per degree of rotation)

Limitations: Exact values of the change of alignment parameters were not reported; for comparison to other studies, the change of TFA and HKA due to limb rotation was extracted form a graph

Jud et al. [24]

Rotation: from 30° ER to 30° IR (10° increments)

Flexion: 0°, 5°, 10°, 15°, 20°, 30°

Others: For each simulation a.p. projections were calculated from the 3D surface model, and the HKA were measured

HKA is not influenced by rotation when the knee is in 0° flexion, but it is by the combination of rotation and flexion. Changes of the HKA are comparable between patients with coronal alignment between 9° varus and 9° valgus:

Rotation and 0° flexion—HKA: Median change referenced to the HKA-baseline 0° (min./max. deviation from the median: -0.8°/0.2°) in 20° ER, 0° (-0.3°/0.0°) in 10° ER 0.1° (-0.3°/0.2°) in 10° IR, 0.2° (-0.9° /0.7°) in 20° IR. Rotation and 10° flexion – HKA: -2.6° (-1.2°/0.3°) in 20° ER, -1.1° (-0.6°/0.2°) in 10° ER, 0.6 (-0.3/0.1) in NR, in 2.3° (-0.4°/0.4°) 10° IR, 3.9° (-0.9°/0.9°) in 20° IR

Limitations: All patients had normal sagittal alignment and femoral torsion; study is based on non-weight-bearing computer simulations

Kannan et al. [25]

Rotation: from 0° to 20° ER (5° increments)

Flexion: 0°, 5°, 10°, 15°, 20°

HKA did not vary > 1° with knee flexion or rotation alone. The combination of both factors can alter the apparent HKA substantially: 10° ER, 5° flexion change of 2°; 10° ER, 10° flexion change of 2°; 20° ER, 10° flexion change of 4°. MPTA and FFA changed by ≤ 1° by flexion or rotation and ≤ 3° in combination of rotation and flexion

Limitations: LLR were not performed in accordance to Paley; IR was not analyzed; only one synthetic sample was analyzed

Kawahara et al. [26]

Rotation: 0°, 5° ER and IR, 10° ER and IR

Others: digitally reconstructed radiography images parallel to the surgical epicondylar axis (neutral rotation; NR), 5° and 10° external rotation (ER) or internal rotation (IR) were reconstructed from preoperative CT. Based on the images, open-wedge HTO was planned aiming for a postoperative WBL of 62.5%. The planned opening angle were measured in each image

The preoperative knee extension angle was − 2.4° ± 3.8°. Weight-bearing line (WBL) changed from 25.6 ± 10.5% in 10°ER, 26.8 ± 9.5% in NR, to 28.5 ± 9.2% in 10°IR. Differences of the WBL between the measurement at 10°, 5° ER, 5°, 10°IR and the neutral position significantly correlated with the knee flexion angle

The analysis of the effect of limb malrotation on preoperative planning of open-wedge HTO showed that the weight-bearing line increased and the medial proximal tibial angle gradually increased from ER to IR, whereas the opening angle gradually decreased. As the opening angle changed only within 0.5° on average, authors concluded that malrotation of the LLR by < 10° does not influence preoperative planning in open-wedge HTO in cases without knee extension deficits. However, knees with > 10° knee extension deficit showed a difference in the measured open angle of > 1° due to rotation

Limitations: Maximum and minimum values of the measured difference were not given. Alignment and flexion deficits were measured using non-weight-bearing CT imaging

Kawakami et al. [27]

Rotation: measured on LLR by superimposing 3D bone models

Flexion: obtained from a lateral view and calculated by the computer software

Others: LLR and 3D Models were obtained using a CT. A computer software was developed to calculate TFA and HKA, its changes due to rotation, to measure knee flexion and simulate knee osteotomy from the 3D bone model. Closed-wedge HTO was simulated with a target of 4° HKA postoperative

The mean limb rotation on radiographs was 7.4 ± 3.9° of IR (range: 8° ER to 14° IR). The mean TFA and HKA on a.p. standing radiographs was 183.5 ± 5.5° (range: 178–205) and -9.2 ± 5.6° (range: -30 to -4). The mean change of HKA was 1.6° ± 1.3° (0.2–4.9) and of the TFA 3.5° ± 2.2° (0.4–8.6°). No significant correlation was found between frontal alignment on AP radiographs and the change in alignment with limb rotation. The change in HKA and FTA with limb rotation increased as the knee flexion increased

Rotation may affect measurement of lower limb alignment for knee osteotomy: With IR and ER by 5°, 10°, and 15° of the osteotomy plane, the FTA changed by 0.2 ± 0.1°, 0.4 ± 0.3°, and 0.8 ± 0.4° and the HKA changed by 0.2 ± 0.1°, 0.4 ± 0.2°, and 0.7 ± 0.3°. HKA and TFA after virtual closed-wedge tibial osteotomy were measured of a case with a rotational error of the osteotomy plane. The proximal cutting surface was reattached to the distal cutting surface. The mean change in TFA was 1.6 ± 0.6°, 3.2 ± 1.1°, and 4.8 ± 1.7° for 5°, 10°, and 15° IR and ER errors in reattaching the osteotomy plane. The mean change in HKA was 0.9° ± 0.5°, 1.9 ± 1.0°, and 2.8 ± 1.5° with an internal and external 5°, 10°, and 15° osteotomy plane rotation

Limitations: transverse images were obtained using a helical CT scanner while the patient was in supine position; only varus knees were included

Kendoff et al. [28]

Rotation: 0°, 5° ER and IR, 10° ER and IR, maximal rotation

Flexion: 0°, 5°, 10°, 20°

Others: The mechanical leg axis and the axis deviations were assessed using a navigation system

In full extension, axis measurement deviations ranging from 0.4° to 4.3° with larger deviations occurring with increasing rotation (IR or ER): mechanical axis deviation with full extension 10° IR 1.5°, 10° ER 2.1°, maximal IR 4.3°, and maximal ER 3.9°. When the knee was flexed, larger deviations were found: 5° flexion with 5°IR led to alignment errors of 3.4°, and with 5°ER to 2.8°. Over- and undercorrection might occur

Limitations: measurement deviation was measured, not the absolute mechanical axis; coronal plane alignment was measured with a navigation system

Khare and Jaramaz [29]

Rotation: 0°, 10° IR, 10° ER

Others: Digitally reconstructed radiographs were obtained from CT scans. Landmarks were manually identified in the radiographs and the 3D CT scans were compared and variations of the positions were measured

Rotation leads to increased errors in certain landmarks (as large as 13.1 mm for the femoral knee center and 13.6 mm for the lateral malleolus). Despite of a large variation in the measurement of landmark points, the estimation of tibial and femoral mechanical axes did not suffer inaccuracy (maximum error of 1.46° for the femoral mechanical axis and 0.66° for the tibial mechanical axis)

Limitations: No weight-bearing condition and radiographs were obtained by digitalized radiographs from a CT scan

Lee et al. [33]

Rotation: 0°, 15° IR an 30° ER of the foot

Foot position in ER shows less varus and IR shows more varus. The weight bearing line shifted laterally in the 30° position and shifted medial in the 15° internal position compared to the neutral position (1.8 mm lateral, 0.2 mm medial in the LLR, and 3.5 mm lateral and 3 mm medial in the local radiograph). Results of the ratio of weight-bearing line/tibial plateau width were similar to those of the weight-bearing line

Limitations: the limb was positioned by the foot position; only standard deviations were presented to report the variation from the mean

Lonner et al. [35]

Rotation: from 20° ER to 25° IR (5° increments)

Flexion: 0°, 10°

ER causes less apparent anatomic valgus and IR pretends more valgus. Even in a well-aligned TKA, limb positioning will alter alignment measurements, making objective evaluation difficult: Rotation and 0° Flexion: 2.6° TFA in 20° ER, 5.7° in 25° IR; Rotation and 10° Flexion: 2.3° in 20° ER, 6.7° in 25° IR. Tibial alignment changed with 10° flexion ranging from -5° varus in 20°ER to 3° valgus in 25°IR

Limitations: mechanical alignment was not measured; non-weight-bearing condition; short-plane radiographs; for comparison to other studies, the change of TFA due to limb rotation was extracted from a graph

Maderbacher et al. [36]

Rotation: rotation errors were calculated by the tibial-fibular overlap using a formula [38]

Flexion: for all radiographs full extension and 10° flexion were assumed

Others: The impact of observed rotational errors was calculated using data of previous studies: for each degree of rotation, the HKA changed in extended knees by 0.0697° and with 10° flexion by 0.0986; femoral and tibial component alignment changed by 0.0824 and 0.0504. Malalignment was defined as a deviation of femoral or tibial component alignment > 2° from 90° or deviation > 3° from neutral HKA (0°)

Limbs were positioned in average in 8.1° ± 9.3° IR (range: 36° IR and 16° ER). Mean measurement errors due to rotation were quite small. However, large ranges of wrong measurements were found indicating significant and clinically relevant measurement errors in some radiographs

Mean differences of HKA between measured alignment before and after a mathematical rotational correction were -0.6° ± 0.6° ranging from -2.5° to 1.1° (0° knee flexion) and -0.8° ± 0.9° ranging from − 3.5° to 1.6° (10° knee flexion). 11 out of 100 patients were wrongly assigned to either mal- or well-aligned, which was defined as HKA within ± 3° varus or valgus

Limitations: see below (Maderbacher et al. [37])

Maderbacher et al. [37]

Rotation: rotation errors were calculated by the fibular overlap using a formula [38]

Flexion: for all radiographs full extension and 10° flexion was assumed

Others: The impact of observed rotational errors on alignment was calculated using data of previous studies: for each degree of rotation AMA changed by 0.0558, mLPFA by 0.1892, mLDFA by 0.0824, mMPTA by 0.0504, mLDTA by 0.2185, HKA in extended knees by 0.0697° and with 10° flexion by 0.0986

Malrotation is regularly present in LLR with average rotation of the limb by -8.0 ± 9.0° IR (range: 29.4° IR and 22.1° ER). Measured alignment parameters before and after rotational correction showed the following mean differences: HKA in full knee extension: 0.6° ± 0.6° (-1.5° to 2.1°); HKA in 10° knee flexion 0.8° ± 0.9° (-2.2° to 2.9°); AMA 0.4° ± 0.5° (-1.2° to 1.6°); mLPFA 1.5° ± 1.7° (-4.2° to 5.6°); mMPTA 0.4° ± 0.5° (-1.1° to 1.5°); mLDFA 0.7° ± 0.7° (-1.8° to 2.4°); mLDTA 1.7° ± 2.0° (-4.8° to 6.4°). As all measured parameters are influenced by malposition, correct limb rotation needs to be verified

Limitations: The formula used to calculate the rotational error can predict lower limb rotation on LLR with a SD of 5°, but with IR the accuracy of calculated rotations decreases; calculated corrections are based on previous studies by Lonner, Jiang and Insall, and Radtke; these values are abstract and vary between mentioned studies. Moreover, these studies were conducted, as outlined by the present review, with small sample sizes; the study did not provide new data regarding the relationship between rotation and alignment deviation

Meijer et al. [40]

Rotation: from 20° ER to 20° IR

Flexion: 0°, 5°, 10°, 15°, 20°

Others: change of the tibiofemoral alignment from 15° varus to 15° valgus (5° increments)

Changes of rotation, flexion, and tibiofemoral alignment in the frontal plane alone had no major impact on the measured HKA (different rotations with 0° varus/valgus and 0° flexion changed between 0.1 and 1.5°). Changes in varus/valgus alignment and rotation with a fully extended knee did not show major differences in HKA (different rotations with 0° flexion and 10° valgus and 10° varus led to a change of HKA by 1.6°–2.1° and -1.7°–0.3). Flexion (20°) combined with rotation with a neutral alignment showed a HKA variation of up to 13.8° (range: -5.2 to 8.6°), a change of rotation and 10° flexion in a neural aligned knee led to a change of HKA between -2.3 and 4.2°. Combining all three factors (15° valgus, 20° flexion) led to maximum HKA variation of up to 16.5° (-10.5–6°). A change of rotation with 10° varus and 10° flexion led to different measured HKA between -2.0–4.9°

Limitations: some values of the change of the alignment parameters were not reported; for comparison to other studies, the change of HKA due to limb rotation was extracted form graphs

Moon et al. [42]

Rotation: was assessed by measuring the deviation of the patellar center inward or outward relative to the midpoint between the femoral condyles

Flexion: the knee extension and flexion was measured on the EOS images

Others: HKA measurement between LLR and EOS was calculated (dHKA) Patients were divided into two subgroups on the sagittal plane (knee flexion and extension group) and into two subgroups based on the axial plane (knee internal and external rotation)

The mean value of the HKA on LLR was 1.2 ± 3.9°, knee flexion angle 5.3 ± 5.3°, and patellar rotation of 4.6 ± 4.0%. Significant correlation was found between dHKA and degree of knee flexion/extension (r = 0.368). No significant correlation was found between dHKA and the axial limb rotation. A significant linear relationship between dHKA and the knee flexion/extension in patients with a patellar rotation more than 3% was found (r² = 0.257)

Authors concluded that the measurement accuracy of coronal alignment on LLR was influences by knee flexion, which was significant in case of small rotation. Therefore, it would be important to check the patella position especially in patients with knee flexion

Limitations: correlations were presented, but no exact values to describe the relationship between rotation and measured HKA; rotation was measured with the patella position

Radtke et al. [47]

Rotation: from 20° ER to 20° IR (5° increments)

Flexion: full extension

Significant effects of rotation on the measured alignment were found: IR simulated tibiofemoral valgus and ER a varus alignment. The alignment parameters in NR were AMA 6.2°, mLPFA 97.5°, mLDFA 88.2°, mMPTA 89.4°, mLDTA 94°. From 20° IR to 20° ER AMA changed from 6.8° to 4.6°, mLPFA from 101.6° to 93.6°, mLDFA 90.6° to 86.8°, mMPTA 90.4° to 88.5°, and mLDTA from 98.9° to 90.5°

Limitations: only one synthetic femur and tibia; HKA or TFA was not measured

Sabharwal et al. [49]

Rotation: gauged as the percentage of proximal fibular overlap and calculated as the percentage of the proximal fibular width that was overlapped by the adjacent tibial metaphyses, at the level of the proximal fibular physis. A large value indicates greater ER

Others: 2 LLR were conducted: one before and the other after ring fixator removal. Difference in the frontal alignment measurement and the limb rotation was measured

The limb tended to be positioned in ER when the radiograph was taken with the circular fixator still attached to the extremity. This tendency for ER of the limb likely contributed to an overestimation of varus alignment. Absolute difference of the MAD between the 2 LLR was 11.5 mm for the treated limb (95%CI with fixator: -3.3 (valgus) to 15.7 mm (varus)); 95%CI without fixator: -6.9 to 10 mm). The mean absolute difference in the fibular overlap between the 2 LLR was 21% (95%CI with fixator: 24% to 48%; 95%CI without fixator: 16% to 33%). Linear regression analysis demonstrated that with an increase of fibular overlap, there was a progressive increase in the magnitude of discrepancy in the measurement of MAD between the 2 LLR

Limitations: the rotation was approximated by the percentage of the proximal fibular overlap

Stricker and Faustgen [55]

Rotation: 0°, 15° ER, 30° ER

Limb malrotation can reduce the validity of HKA and TFA progressively. HKA: -27.7° in NR and increased by 2.4° in 15° ER and 3.0° in 30° ER; TFA: -24.8° in NR and increased by 3.0° in 15° ER and 6.7° in 30° ER. For MDA (average 12.8° in 0° rotation according to Levine-Drennan) unpredictable variability was noted during ER, which can lead to misdiagnosis and mistreatment of Blount’s disease

Limitations: the limb positioning was controlled with ultrasound; only standard deviations were presented to report the variation from the mean

Swanson et al. [56]

Rotation: from 20° ER to 20° IR (10° increments)

Others: Three deformed limbs: A 18° valgus, B 7° varus and C 7° valgus (TFA). Each model had 5 radiographs in each position

With a graphic software, high tibial osteotomies were simulated for model B. Osteotomies aimed to achieve 8° valgus

The effect of rotation on measured TFA was more sensitive in varus and valgus deformed limbs than in a neutral limb. Severe valgus deformities are more sensitive to the effects of ER; severe varus deformities are more sensitive to effects of IR. TFA of model A: 13.6° in 20° ER, 16.7° in 10° ER, 18.7° valgus in NR, 20.2° in 10° IR, 21.4° in 20° IR. Model B: -7.8° in 20° ER, -7.4° in 10° ER, -6.4° in NR., -5.3° in 10° IR, -3.7° in 20° IR. Model C: 4.6° in 20° ER, 6.3° in 10° ER, 6.3° in NR, 6.9° in 10° IR, 7.3° in 20° IR. AMA varied less than 1° in all three models between 20° IR and ER

A simulated osteotomy based on the measured TFA of model B in 20° IR an undercorrection occurred. Ideal correction of a varus deformity was achieved when the osteotomy was simulated based on LRR with the limb in the neutral position. Based on the measured TFA in 20°ER, an overcorrection was observed

Limitations: axial rotation was assessed with the help of a goniometer

Thelen et al. [58]

Rotation: from 20° ER to 20° IR (5° increments)

Flexion: 1 bone in 0°, 1 bone in 9°, 1 bone in 18°

Others: Biplanar slot scanning system (EOS) took an a.p. and a lateral radiograph of the models (5° valgus) for each rotation position

The impact of rotation on the HKA is relatively small in the absence of sagittal knee angulation. An axial rotation of 10° (IR or ER) resulted in a measurement error (defined as difference between the measured HKA and the actual model alignment of 5°) in the 2D a.p. view of 0.4°, 1.9°, and 3.1° and an axial rotation of 20° (IR or ER) resulted in an average measurement error of 1.4°, 4.7°, and 6.8° with 0°, 9°, and 18° of knee flexion. Summarizing the measurements of 3 observers, HKA varied from 3.8°-6.4° (model 0° flexion), 0.3–9.7° (model 9° flexion) and -0.7 – 11.8° (model 18° flexion)

Limitations: only standard deviations were presented to report the variation from the mean; for comparison to other studies, the change of HKA due to limb rotation was extracted form a graph

Wright et al. [63]

Rotation: from 20° ER to 20° IR (10° increments)

Flexion: 0°, 20°, 30°

Little overall difference between the rotation groups were found. The reliability of lower limb alignment measured from LLR is satisfactory. The radiographer positioned the limb within 3.1° ± 2.7° from the neutral position

Rotation and 0° Flexion of the first limb: 2.0° in 20° ER, 1.7° in 10° ER, 2.0° in NR, 2.3° in 10° IR, 2.0° in 20° IR. Rotation and 0° Flexion of the second limb: 7.0° in 20° ER, 3.3° in 10° ER, 1.0° in NR, -0.7° in 10° IR, -2.0° in 20° IR. Rotation and 20° Flexion—TFA: 2.0° in 20° ER, 2.0° in NR, 0.7° in 20° IR of the first limb; 2.7° in 20° ER, 4.6° in NR, 4.0° in 20° IR of the second limb

Limitations: different findings between the limbs; only two knee amputates were analyzed; only anatomical measurements were analyzed

Yoo et al. [64]

Rotation: assessed by measuring the fibula/tibia overlap, as well as the tibial and femoral component rotation from AP radiographs conducted by EOS imaging system

Flexion: Flexion was measured from lateral EOS radiographs

Others: correlations referring to the HKA angle were calculated

There was a significant correlation for the fibular overlap (r = 0.292)/the femoral component rotation (r = 0.317) and the HKA angle. Moreover, combined rotation and knee flexion had a greater effect on the coronal alignment than flexion/rotation alone. This was more frequently observed during the early period after TKA

Limitations: Rotation was only described by the fibular overlap and the component position. Relationship between rotation/flexion was expressed by calculating correlation coefficients

Descriptions of the included studies were summarized, simplified, and focused on the aim of the present systematic review (negative values of HKA and TFA indicate varus alignment

mLDFA mechanical lateral distal femoral angle, MPTA medial proximal tibial angle, wbLLR weight-bearing long-leg radiographs, LLR non-weight-bearing long-leg radiographs, MPTA medial proximal tibial angle, aLDFA anatomic lateral distal femoral angle, AMA anatomical mechanical angle of the femur, MAD mechanical axis deviation, FFA frontal femoral angle, MDA proximal tibial metaphyseal–diaphyseal angle, JLCA joint line convergence angle, bowing angle of the femur (angle between the anatomic femoral axis and the proximal femoral shaft), WBL distance from medial edge of tibia plateau to mechanical limb axis in mm, NR neutral rotation, IR internal rotation, ER external rotation